CN117115728A - Risk identification method and system applied to field operation of transformer substation - Google Patents

Risk identification method and system applied to field operation of transformer substation Download PDF

Info

Publication number
CN117115728A
CN117115728A CN202310857545.3A CN202310857545A CN117115728A CN 117115728 A CN117115728 A CN 117115728A CN 202310857545 A CN202310857545 A CN 202310857545A CN 117115728 A CN117115728 A CN 117115728A
Authority
CN
China
Prior art keywords
point cloud
point
dimensional
data
real
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310857545.3A
Other languages
Chinese (zh)
Inventor
张祥全
单长旺
王利平
周华良
聂江龙
卢璐
贺洲强
苏战涛
陈钊
狄磊
马宏忠
王晶
孙瀚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
STATE GRID GASU ELECTRIC POWER RESEARCH INSTITUTE
State Grid Gansu Electric Power Co Ltd
State Grid Electric Power Research Institute
Original Assignee
STATE GRID GASU ELECTRIC POWER RESEARCH INSTITUTE
State Grid Gansu Electric Power Co Ltd
State Grid Electric Power Research Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by STATE GRID GASU ELECTRIC POWER RESEARCH INSTITUTE, State Grid Gansu Electric Power Co Ltd, State Grid Electric Power Research Institute filed Critical STATE GRID GASU ELECTRIC POWER RESEARCH INSTITUTE
Priority to CN202310857545.3A priority Critical patent/CN117115728A/en
Publication of CN117115728A publication Critical patent/CN117115728A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G06Q50/265—Personal security, identity or safety
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition
    • G06V10/16—Image acquisition using multiple overlapping images; Image stitching
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/64—Three-dimensional objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G08B21/24—Reminder alarms, e.g. anti-loss alarms
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B3/00—Audible signalling systems; Audible personal calling systems
    • G08B3/10—Audible signalling systems; Audible personal calling systems using electric transmission; using electromagnetic transmission
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Economics (AREA)
  • Medical Informatics (AREA)
  • Tourism & Hospitality (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Computational Linguistics (AREA)
  • General Business, Economics & Management (AREA)
  • Primary Health Care (AREA)
  • Marketing (AREA)
  • Image Analysis (AREA)
  • Emergency Management (AREA)
  • Electromagnetism (AREA)
  • Social Psychology (AREA)
  • Human Computer Interaction (AREA)
  • Psychiatry (AREA)
  • Water Supply & Treatment (AREA)
  • Computer Security & Cryptography (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Public Health (AREA)

Abstract

The invention discloses a risk identification method applied to field operation of a transformer substation, which comprises the following steps: clustering point clouds and semantic segmentation are carried out on the three-dimensional scenic spot cloud map; loading a three-dimensional point cloud map, importing a working ticket, and arranging a substation safety control terminal and a laser radar module; clustering and semantic segmentation are carried out on the point clouds in the working area, and fusion matching of the three-dimensional point cloud map and the point cloud data is completed; real-time monitoring is carried out through real-time positioning of staff, and identification of personnel identity information and illegal behaviors is completed; and identifying the semantic point cloud, matching a risk identification library for deep learning, and identifying and alarming personnel risk behaviors. The invention also discloses a risk identification system applied to the field operation of the transformer substation, which comprises a device and a module matched with the method. The invention has the advantages of high response speed, wide coverage, video linkage auxiliary monitoring, on-site real-time voice alarm prompt, full operation flow video retention and the like.

Description

Risk identification method and system applied to field operation of transformer substation
Technical Field
The invention belongs to the technical field of transformer substation safety operation, and particularly relates to a risk identification method and system applied to transformer substation field operation.
Background
At present, along with the upgrading and accelerating of the construction of a novel power system, new equipment, new technology, new system and new operation management mode are continuously popularized and applied in power transmission and transformation engineering, the risk identification difficulty is further improved, the existing operation management and control force is insufficient, the monitoring capability is limited, new safety risk points are continuously appeared, new challenges are provided for the operation capability and behavior standardization level of field operators, and the safety production situation is still severe.
The safety monitoring mode of the power transmission and transformation operation site mainly comprises the steps of staring at person monitoring, carrying out remote monitoring and intelligent identification by utilizing a video camera, and supervising the working route of an operator by wearing a belt positioning terminal, wherein the staring at person monitoring mode is limited by the modes of uneven personnel quality, personnel carding shortage and wide working range, a plurality of potential safety hazards cannot be effectively identified and prompted, and the mode exists in the case of a safety event; the traditional two-dimensional image recognition technology is difficult to capture accurate spatial position information of personnel, equipment and environment, and the foreground and the background of an electric power scene cannot be autonomously distinguished, so that intelligent promotion of power grid engineering safety risk recognition and operation management and control is restricted; wearable positioning terminals such as intelligent helmets, intelligent tools and the like with GPS positioning functions are affected by wearing modes, application scenes and the like, general precision deviation is between 0.5 meter and 1.5 meter, and high-precision noninductive positioning of on-site operators is difficult to achieve.
Based on the defects of the safety monitoring mode, in the prior art, the Chinese patent application number is 2020110519231 and is named as a substation site operation safety early warning method, and a three-dimensional point cloud scanning method and an image monitoring method are combined to treat specific problems, so that the method has the characteristics of modularization, high accuracy, wide detection area, comprehensive supervision and substation site operation data informatization; however, the method combines three-dimensional point cloud and image monitoring, so that the calculation amount of the whole system is increased, and the response speed is slow in actual use, so that the risk of field operation is increased; secondly, fusion matching of the three-dimensional map and the real-time point cloud is mainly realized through a camera, and the matching error between the map and the point cloud is large, so that the positioning precision and the abnormality detection precision are poor in field operation.
Disclosure of Invention
The invention aims to: the invention aims to provide a risk identification method with good safety and high positioning precision for field operation of a transformer substation; another object of the present invention is to provide a risk identification system applied to field operations of a substation.
The technical scheme is as follows: the risk identification method applied to the field operation of the transformer substation comprises the following steps:
carrying out three-dimensional real-scene point cloud scanning on a scene of the transformer substation, and collecting three-dimensional point data and real-scene photo data of the transformer substation;
converting the collected three-dimensional point position data and live-action photo data of the transformer substation into a three-dimensional live-action point cloud map of the transformer substation;
clustering point clouds and semantic segmentation are carried out on the three-dimensional scenic spot cloud map;
loading a three-dimensional point cloud map, importing a working ticket, and arranging a substation safety control terminal and a laser radar module;
clustering and semantic segmentation are carried out on the point clouds in the working area, and fusion matching of the three-dimensional point cloud map and the point cloud data is completed;
real-time monitoring is carried out through real-time positioning of staff, and identification of personnel identity information and illegal behaviors is completed;
and identifying the semantic point cloud, matching a risk identification library for deep learning, and identifying and alarming personnel risk behaviors.
The method comprises the steps of converting the collected three-dimensional point data and live-action photo data of the transformer substation into a three-dimensional live-action point cloud map of the transformer substation, and converting the collected three-dimensional point cloud data of the transformer substation into the three-dimensional live-action point cloud map of the transformer substation through data analysis, track analysis, coordinate conversion, point cloud generation, point cloud coloring and map release.
The method comprises the steps of carrying out point cloud clustering and semantic segmentation on a three-dimensional scenic spot cloud map, wherein the step of carrying out point cloud clustering and semantic segmentation on the three-dimensional scenic spot cloud map comprises the steps of segmenting each device point cloud through the point cloud clustering, and completing automatic labeling of the device semantic point cloud through deep learning to match typical point clouds of each device; the clustering of the point clouds comprises the steps of preprocessing, filtering noise points through an outlier point cloud filtering method, and reducing the influence of the noise point clouds on equipment clustering segmentation; filtering the ground point cloud, detecting the ground through a random sampling execution algorithm, and distinguishing the detected ground data from the equipment point cloud to be clustered; and finally, accumulating the point clouds of the equipment, clustering and segmenting the point clouds to be clustered by using an European clustering algorithm based on nearest neighbor search, and matching with the point cloud information of typical equipment in a database to realize the clustering of data and complete the segmentation of semantic point clouds of the point cloud equipment.
Wherein the random sampling execution algorithm detecting the ground comprises setting the iteration number of the algorithm as I and the distance error threshold as delta T 1 The total point number of the point cloud is N, 3 points are randomly selected to form a ground to be fitted at the beginning, and three-dimensional coordinates of the 3 points are set as (X 1 ，Y 1 ，Z 1 )、(X 2 ，Y 2 ，Z 2 )、(X 3 ，Y 3 ，Z 3 ) The fitted planar model is:
Ax+By+Cz+D＝0；
wherein:
A＝(Y 2 -Y 1 )(Z 3 -Z 1 )-(Z 2 -Z 1 )(Y 3 -Y 1 )；
B＝(Z 2 -Z 1 )(X 3 -X 1 )-(X 2 -X 1 )(Z 3 -Z 1 )；
C＝(X 2 -X 1 )(Y 3 -Y 1 )-(Y 2 -Y 1 )(X 3 -X 1 )；
D＝-(AX 1 +BY 1 +CZ 1 )；
then any point (X) 0 ，Y 0 ，Z 0 ) The distance L to this plane is as follows:
when the distance L between a certain point and the assumed plane is less than or equal to delta T 1 When the model is in the model, the point is the inner point of the model; traversing other N-3 points except the initial sampling 3 points in sequence, and recording the number of the inner points of the model; randomly sampling 3 points to construct a plane model, and obtaining the number of internal points of the model according to the same method; iterating for I times according to the random sampling method; the probability of producing reasonable results increases with the increase of the iteration times; finally, voting based on the number of the internal points of each model, and selecting the ground model with the largest number of the internal points as the best fitting result; finally, the division of the ground point cloud and the point cloud with the clusters is finished, and only the point cloud with the clusters is required to be processed in the follow-up process, so that the calculation amount of the clusters is greatly reduced; when point cloud clustering is performed using the European clustering algorithm of nearest neighbor search, it is assumed that the node coordinates to be searched are (X n ，Y n ，Z n ) The coordinates of the target point are (X t ，Y t ，Z t ) And (3) making:
Δx＝X n -X t ；
Δy＝Y n -Y t ；
Δz＝Z n -Z t ；
if the 3 conditions of the I delta x I delta T, the I delta y I delta T and the I delta z I delta T are simultaneously met, calculating the Euclidean distance D between the detection point and the target point, wherein the calculation formula is as follows:
if D is less than or equal to delta T, the node is considered to be the nearest neighbor of the target point; searching all the points according to the method until all the nearest neighbor points meeting the distance error threshold are searched; recording the searched nearest neighbors of the target point in a set, and matching the point cloud data of the set with the point cloud data in a database to obtain the equipment attribute of the clustered point cloud.
The method comprises the steps that a three-dimensional point cloud map is loaded, a working ticket is imported, a substation safety management and control terminal and a laser radar module are arranged, the working ticket is acquired and automatically analyzed, working members and working content information of the working ticket are automatically read through a mode of accessing a PMS (permanent magnet synchronous system) to automatically import the working ticket or the working ticket imported into an electronic version, an electronic fence is automatically produced, and virtual security measure arrangement is completed; the laser radar module is arranged in such a way that a fixed-point three-dimensional laser radar module is arranged at the edge position of a working area, the laser radar realizes communication with main service equipment in a battery power supply and wireless communication mode, and the fixed-point three-dimensional laser radar module comprises a three-dimensional laser radar, an omnidirectional holder, voice intercom, voice broadcasting, wireless communication, a battery, power management and an edge processor; the transformer substation safety control terminal is a movable edge control device integrating laser radar network access, video access, data processing and control and communication on pairs, and functions of laser radar data analysis, video linkage and analysis, on-site alarm communication, remote alarm pushing and the like are realized.
Clustering and semantic segmentation are carried out on the point clouds in the working area, fusion matching of the three-dimensional point cloud map and the point cloud data is completed, multi-machine calibration of the laser radar is included, and the three-dimensional map is matched with the laser radar real-time data; the laser radar collects real-time point cloud data of a working area and provides original data for subsequent functional analysis; the three-dimensional map and the real-time point cloud data are subjected to water level fusion matching, the calibrated laser radar data and the three-dimensional map data are in the same coordinate system, the real-time data collected by the laser point cloud can be matched with the three-dimensional map, and the real-time point cloud of the working area is displayed.
The three-dimensional map and real-time point cloud data water level fusion matching comprises the steps of firstly constructing two point sets P and Q, wherein P= { Pi, i=1, 2,3, & gt, k }, Q= { Qi, i=1, 2,3, & gt, n }; placing real-time points in a point set P, storing point cloud data of a three-dimensional map in a point set Q, wherein the number of the two point sets is not necessarily equal, and the conventional condition k is larger than n; the matching process is to find a rotation matrix R and a translation matrix T through rotation and translation of a coordinate system of Q, minimize the distance between corresponding points in a point set P in the point set Q, and after registration, the relationship between each point in P and Q is as follows:
P i ＝RQ i +T；
then solving a rotation matrix R and a translation matrix based on a quaternion method, and erecting a quaternion q of a unit R ＝[q 0 ，q 1 ，q 2 ，q 3 ] T WhereinAnd q 0 > 0, then the rotation matrix R is as follows:
the translation matrix is defined as q T ＝[q 4 ，q T5 ，q 6 ] T The registration vector formed by the rotation matrix and the translation matrix is After the center of gravity of the point sets P and Q is obtained and R and T are finally obtained, the Euclidean distance between the point sets P and Q can be obtained as follows:
wherein L is 2 The smaller the value is, the smaller the matching error of the two point sets is, the higher the point set identity is, and the matching of the point cloud is completed.
The real-time positioning of the staff comprises the steps of analyzing the point cloud data of the staff by using a background subtraction method based on the point cloud, filtering the discrete point cloud out of the ground and equipment, and obtaining a moving point cloud cluster; clustering the point cloud clusters, obtaining the boundary of the point cloud by adopting a rolling method, then carrying out three-dimensional projection on the point cloud, and extracting point cloud characteristic information to finish target classification and point cloud skeleton information; and carrying out real-time positioning analysis on the point cloud through the point cloud type and the skeleton information to obtain real-time position data of the point cloud of the staff.
The identification of the semantic point cloud, the matching of a risk identification library of deep learning, the identification and alarming of personnel risk behaviors comprise video tracking and image identification, the real-time position of the personnel is obtained through the video tracking and image identification, the identification based on the face information is carried out by combining with an image identification algorithm, and the identification of illegal and illegal behaviors is carried out; based on the personnel behavior recognition of the semantic point cloud, the distance information of the personnel and the equipment is obtained by simulating the point cloud boundary through an ellipsoid model, the distance information is used for judging whether the safety distance element is installed or not, whether the behaviors of out-of-limit and wrong entering interval exist or not is judged by combining the electronic fence information, and whether the current agent has the illegal climbing behavior or not is judged according to the gravity center height; based on risk discrimination of semantic point cloud, backbone information of the semantic point cloud is fused with joint information of personnel, trend analysis of the semantic point cloud is carried out by matching the current frame point cloud with data in the previous history 10 frames, trend discrimination is carried out on the personnel, timely voice prompt is carried out on the action with potential safety hazard, and the personnel is timely reminded.
A risk identification system for use in substation field operation, comprising:
the three-dimensional scenic spot cloud scanning module is used for carrying out three-dimensional scenic spot cloud scanning on a scene of the transformer substation and collecting three-dimensional point position data and scenic photograph data of the transformer substation;
the three-dimensional real scenic spot cloud data processing module is used for converting the collected three-dimensional point data and real scenic photograph data of the transformer substation into a three-dimensional real scenic spot cloud map of the transformer substation, and clustering and semantic segmentation of the point cloud are carried out on the three-dimensional real scenic spot cloud map;
the multi-machine radar deployment and calibration module is used for loading a three-dimensional point cloud map, importing a working ticket and arranging a substation safety control terminal and a laser radar module;
the real-time point cloud acquisition module comprises a module for clustering and semantically dividing the point cloud in the working area to finish fusion matching of the three-dimensional point cloud map and the point cloud data,
the non-inductive personnel positioning module is used for monitoring personnel identity information and identifying illegal behaviors in real time through real-time positioning of the personnel;
the personnel behavior risk recognition module is used for recognizing the semantic point cloud and matching a risk recognition library for deep learning;
and the video linkage and voice alarm module is used for identifying and alarming personnel risk behaviors.
The beneficial effects are that: compared with the prior art, the invention has the following remarkable progress: the invention utilizes the three-dimensional laser point cloud scanning and detection technology and combines the image recognition technology, and realizes centimeter-level positioning accuracy and abnormal behavior detection by constructing the three-dimensional space point cloud of the working area and detecting the working area point cloud in an omnibearing manner, thereby having the advantages of high response speed, wide coverage range, video linkage auxiliary monitoring, on-site real-time voice alarm prompt and full operation flow video retention; meanwhile, the preprocessing of the point cloud reduces the operation amount of the system and further quickens the response speed of the system; secondly, calculating Euclidean distance in the process of fusion matching of the three-dimensional map and the real-time point cloud, and improving the matching precision of the map and the real-time point cloud; the method has great help to solve the illegal or unsafe behaviors of the power transmission and transformation station, such as wrong entering interval, out-of-range operation, unreasonable use of large tools and instruments, and the like, and improves the safety of field operators and equipment.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Detailed Description
The flow diagram of the invention is shown in fig. 1, and the risk identification method applied to the field operation of the transformer substation comprises the following steps: the method comprises the steps of three-dimensional real scenic spot cloud scanning of a transformer substation scene, three-dimensional real scenic spot cloud data processing, work ticket importing and electronic fence deployment, multi-machine radar deployment and calibration, real-time point cloud acquisition and processing, non-inductive personnel positioning, personnel behavior risk identification, safety control module, video linkage and voice alarm.
The three-dimensional real-scene point cloud scanning is carried out on a scene of a transformer substation, three-dimensional point position data and real-scene photo data of objects such as equipment and buildings in the transformer substation are obtained, the three-dimensional real-scene point cloud scanning comprises a three-dimensional laser radar, a panoramic camera, a high-precision inertial navigation device, a collection and storage industrial personal computer, a motion mechanism or a bearing mechanism and a battery power supply module, and the modules are fixedly installed on the bearing mechanism, so that data collection is facilitated. Because three-dimensional scenic spot cloud scanning equipment relates to a plurality of sensors, each equipment passes through chargeable battery power supply and accomplishes data communication with acquisition system, and the staff carries data acquisition equipment, accomplishes full coverage road walking in the station, with the collection of the interior raw data of station and store the industrial computer of carrying on in, include: three-dimensional laser point cloud data, panoramic camera image data, high-precision inertial navigation position information, and the like.
The calibration and conversion work of the equipment coordinate system are carried out through the automatic calibration function of the acquisition software system, and the coordinate systems of the sensors are unified; starting acquisition software, actively connecting a laser radar, a panoramic camera and high-precision high-end equipment, triggering an acquisition button after connection is correct, starting data acquisition, carrying the acquisition equipment to walk in a transformer substation to perform data acquisition, and covering passable roads as much as possible when no work is performed in the transformer substation during data acquisition to ensure the acquisition effect and ensure the data acquisition precision; after the data acquisition is completed, stopping the data acquisition work, storing and exporting the acquired data, and after all the work is completed, exiting the acquisition software and closing the equipment.
The real-time point cloud acquisition and processing comprises the steps of importing acquired substation data into a three-dimensional real-scene point cloud processing software system, and converting the acquired substation three-dimensional point cloud data into a substation three-dimensional real-scene point cloud map through data analysis, track analysis, coordinate conversion, point cloud generation, point cloud coloring, map release and other works.
The data analysis is to analyze the point cloud file, the live-action file and the inertial navigation file in the engineering file into temporary file formats required by data processing; then track analysis is carried out, and track information of equipment walking is obtained through loop detection; matching track information with a coordinate system, completing conversion of the coordinate system, after unifying the coordinate system by sensor data, producing point cloud with coordinate (x, y, z) information, coloring the point cloud to realize color point cloud information of equipment and facilities in a transformer substation, calculating and generating a three-dimensional point cloud map of the transformer substation by adopting a laser SLAM technology, and issuing a visualized three-dimensional point cloud map.
The point cloud segmentation and identification are to automatically acquire semantic point clouds of equipment in the transformer substation through clustering analysis of the point clouds.
The three-dimensional point cloud map generated automatically is discrete point cloud data, and the point cloud information of equipment cannot be automatically screened out, because equipment and layout in a transformer substation have certain rules and can be circulated, the point cloud of each equipment is segmented through clustering of the point clouds, and the automatic labeling of the semantic point cloud of the equipment is completed through deep learning of typical point clouds matched with each equipment.
The point cloud clustering is firstly preprocessed, noise points are filtered through an outlier point cloud filtering method, and the influence of the noise point cloud on equipment clustering segmentation is reduced; filtering the ground point cloud, wherein the equipment is above the ground, a large amount of point cloud data on the ground is filtered to reduce the operation amount, the ground is detected through a random sampling execution algorithm, and the detected ground data is distinguished from the equipment point cloud to be clustered; and finally, accumulating the point clouds of the equipment, clustering and segmenting the point clouds to be clustered by using an European clustering algorithm based on nearest neighbor search, and matching with the point cloud information of typical equipment in a database to realize the clustering of data and complete the segmentation of semantic point clouds of the point cloud equipment.
When the machine sampling execution algorithm detects the ground, firstly, the iteration number of the algorithm is set as I, and the distance error threshold value is deltaT 1 The total point number of the point cloud is N, 3 points are randomly selected to form a ground to be fitted at the beginning, and three-dimensional coordinates of the 3 points are set as (X 1 ，Y 1 ，Z 1 )、(X 2 ，Y 2 ，Z 2 )、(X 3 ，Y 3 ，Z 3 ) The fitted planar model is shown in formula (1):
Ax+By+Cz+D＝0 (1)
wherein:
A＝(Y 2 -Y 1 )(Z 3 -Z 1 )-(Z 2 -Z 1 )(Y 3 -Y 1 )；
B＝(Z 2 -Z 1 )(X 3 -X 1 )-(X 2 -X 1 )(Z 3 -Z 1 )；
C＝(X 2 -X 1 )(Y 3 -Y 1 )-(Y 2 -Y 1 )(X 3 -X 1 )；
D＝-(AX 1 +BY 1 +CZ 1 )；
then any point (X) 0 ，Y 0 ，Z 0 ) The distance L to this plane is shown in formula (2):
when the distance L between a certain point and the assumed plane is less than or equal to delta T 1 When it is, then thisIs the model interior point. And traversing other N-3 points except the initial sampled 3 points in sequence, and recording the number of the inner points of the model.
And randomly sampling 3 points to construct a plane model, and obtaining the number of inner points of the model according to the same method. The method of random sampling is iterated 1 time. The probability of producing a reasonable result increases with increasing iteration number. And finally, voting based on the number of the internal points of each model, and selecting the ground model with the largest number of the internal points as the best fitting result. Finally, the division of the ground point cloud and the point cloud with the clusters is finished, and only the point cloud with the clusters is required to be processed in the follow-up process, so that the calculation amount of the clusters is greatly reduced;
when point cloud clustering is performed using the European clustering algorithm of nearest neighbor search, it is assumed that the node coordinates to be searched are (X n ，Y n ，Z n ) The coordinates of the target point are (X t ，Y t ，Z t ) Order-making
Δx＝X n -X t ； (3)
Δy＝Y n -Y t ； (4)
Δz＝Z n -Z t ； (5)
If the 3 conditions of deltax delta T delta y delta T delta z delta T are met at the same time, calculating the Euclidean distance D between the detection point and the target point, wherein the calculation formula is
If D is less than or equal to DeltaT, the node is considered to be the nearest neighbor of the target point. And searching all the points according to the method until all the nearest neighbor points meeting the distance error threshold are searched. Recording the searched nearest neighbors of the target point in a set, and matching the point cloud data of the set with the point cloud data in a database to obtain the equipment attribute of the clustered point cloud.
The steps can be completed once, the point cloud data is updated periodically after the equipment area is changed, and the follow-up steps are re-executed according to the working content of each time.
The method comprises the steps of on-site deployment of the three-dimensional laser radar equipment, firstly obtaining working contents according to work ticket information, and then deploying a safety control edge terminal and a laser radar monitoring module.
The working ticket is acquired and automatically analyzed, the working member and the working content information of the working ticket are automatically read by accessing the PMS to automatically import the working ticket or the working ticket imported into the electronic version, the electronic fence is automatically produced, and the arrangement of the virtual security measure is completed.
The three-dimensional laser radar monitoring module is arranged at the edge position of the working area, preferably at four corners of the working area, the laser radar adopts a battery power supply and wireless communication mode to realize communication with the main service equipment, and the three-dimensional laser radar monitoring module comprises a three-dimensional laser radar, an omnidirectional cloud deck, voice intercom, voice broadcasting, wireless communication, a battery, power management and an edge processor.
The transformer substation safety control terminal is a movable edge control device integrating laser radar network access, video access, data processing and control and communication on pairs, and functions of laser radar data analysis, video linkage and analysis, on-site alarm communication, remote alarm pushing and the like are realized.
The laser radar is preferably a three-dimensional laser radar with more than 16 lines to collect site point clouds, the principle is that three-dimensional distance information is obtained by transmitting and receiving laser and utilizing a TOF principle, the distance information is combined into single-frame point cloud data to be issued through an edge processing node, the detection distance can reach 100m, and the laser radar is horizontally installed on an upright post with the height of 1.5 m in order to follow the principle that the field of view and blind areas of the laser radar are fully utilized as small as possible.
The multi-machine calibration of the three-dimensional laser radar is realized by carrying out position calibration on a three-dimensional map according to the deployment position of the laser radar through a scanned three-dimensional map of a transformer substation, and matching the three-dimensional map with real-time data of the laser radar.
The three-dimensional point cloud visual map is logged in, a rough position is selected in the three-dimensional point cloud map to serve as an initial value of point cloud matching according to the actual deployment position of the laser radar, point cloud data actually acquired by the laser radar are matched with the point cloud map, and a nonlinear optimization algorithm is used for solving the position coordinates of the laser radar, so that the position coordinates of the laser radar on the three-dimensional point cloud map are accurately calibrated.
Because of the existence of a plurality of laser radars, the calibration of each laser radar needs to be completed in sequence, each laser radar mutually supplements a shielding area, and no dead angle fully covers a working area.
The laser radar collects real-time point cloud data of the working area and provides original data for subsequent functional analysis.
Because the number of the field laser radars is large, laser data of each radar is not distinguished, and equipment id numbers are attached to the point cloud data.
The three-dimensional map and the real-time point cloud are fused and matched, and the calibrated laser radar data and the data of the three-dimensional map are in the same coordinate system, so that the real-time data collected by the laser point cloud can be matched with the three-dimensional map, and the real-time point cloud of the working area is displayed.
First two point sets P and Q are constructed, where p= { Pi, i=1, 2,3,..k }, q= { Qi, i=1, 2,3,., n }. The real-time points are put in the point set P, the point cloud data of the three-dimensional map are stored in the point set Q, the number of the two point sets is not necessarily equal, and the conventional case k is larger than n. The matching process is to find a rotation matrix R and a translation matrix T through rotation and translation of a coordinate system of Q, minimize the distance between corresponding points in a point set P in the point set Q, and after registration, the relationship between each point in P and Q is as follows:
P i ＝RQ i +T； (7)
then solving a rotation matrix R and a translation matrix based on a quaternion method, and erecting a quaternion q of a unit R ＝[q 0 ，q 1 ，q 2 ，q 3 ] T WhereinAnd q 0 > 0, then the rotation matrix R is as follows:
let the definition of the translation matrix be q T ＝[q 4 ，q T5 ，q 6 ] T The registration vector formed by the rotation matrix and the translation matrix isAnd then the barycenter of the point sets P and Q is respectively obtained, and finally R and T are obtained, so that the Euclidean distance between the point sets P and Q can be obtained as follows:
wherein L is 2 The smaller the value is, the smaller the matching error of the two point sets is, the higher the point set identity is, and the matching of the point cloud is completed.
The background filtering of the real-time point cloud is to filter the background data of the real-time point cloud according to the point cloud information of the three-dimensional map and the real-time point cloud information, and the real-time point cloud computing amount is reduced and the computing speed is improved by wrapping the point clouds such as the ground, the field operation equipment and the upright post.
And analyzing the point cloud data of the staff by using a background subtraction method based on the point cloud, and filtering the discrete point cloud out of the ground and equipment after the point cloud matching is completed to obtain a moving point cloud cluster.
Clustering and analyzing the point cloud clusters, and completing the point cloud fitting of personnel or tools in the working area by clustering the filtered point cloud clusters to obtain a backbone network of the point cloud clusters.
Clustering the point cloud clusters, obtaining the boundary of the point cloud by adopting a rolling method, then carrying out three-dimensional projection on the point cloud, and providing point cloud characteristic information to finish target classification and point cloud skeleton information.
And carrying out real-time positioning through the point cloud, and carrying out real-time positioning analysis on the point cloud through the type and skeleton information of the point cloud to obtain the position data of the point cloud.
When in real-time positioning, personnel positioning detection is executed once every 100 milliseconds, the module collects laser radar data within 100 milliseconds and is spliced into a single-frame point cloud, and personnel position data is returned through the steps.
Video tracking and image recognition, and combining an image recognition algorithm to perform identity recognition based on face information and recognize illegal and illegal behaviors such as non-wearing safety helmet, non-wearing tool, smoking, low hanging and high use of safety belt.
And (3) based on the personnel behavior recognition of the semantic point cloud, simulating a point cloud boundary through an ellipsoid model, obtaining distance information of a worker and equipment, judging whether a safety distance is required to be installed, judging whether behaviors such as out-of-limit and wrong entering interval exist or not through combining electronic fence information, and judging whether the current agent has illegal climbing behaviors or not according to the gravity center height.
Based on risk discrimination of semantic point cloud, backbone information of the semantic point cloud is fused with joint information of personnel, trend analysis of the semantic point cloud is carried out by matching the current frame point cloud with data in the previous history 10 frames, trend discrimination is carried out on the personnel, timely voice prompt is carried out on the action with potential safety hazard, and the personnel is timely reminded.
And (3) carrying out on-site alarming and video linkage snapshot of abnormal behaviors, completing abnormal behavior judgment of staff through steps 10-13, carrying out on-site voice alarming after the abnormal behaviors are identified through point cloud data of one frame every 100ms until the alarming disappears or is manually reset, stopping alarming, splicing videos during alarming by colleagues, and restoring and recording a real alarming process. And finally uploading the alarm information to a higher-level security management and control platform.
And finishing work. When all works are finished or the work ticket is checked, the laser radar device and the substation safety control terminal are required to be retracted, and the work receiving operation is performed to form a work closed loop.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. The scheme in the embodiment of the invention can be realized by adopting various computer languages, such as object-oriented programming language Java, an transliteration script language JavaScript and the like.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (10)

1. A risk identification method applied to field operation of a transformer substation is characterized by comprising the following steps of: the method comprises the following steps:
carrying out three-dimensional real-scene point cloud scanning on a scene of the transformer substation, and collecting three-dimensional point data and real-scene photo data of the transformer substation;
converting the collected three-dimensional point position data and live-action photo data of the transformer substation into a three-dimensional live-action point cloud map of the transformer substation;
clustering point clouds and semantic segmentation are carried out on the three-dimensional scenic spot cloud map;
loading a three-dimensional point cloud map, importing a working ticket, and arranging a substation safety control terminal and a laser radar module;
clustering and semantic segmentation are carried out on the point clouds in the working area, and fusion matching of the three-dimensional point cloud map and the point cloud data is completed;
real-time monitoring is carried out through real-time positioning of staff, and identification of personnel identity information and illegal behaviors is completed;
and identifying the semantic point cloud, matching a risk identification library for deep learning, and identifying and alarming personnel risk behaviors.
2. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the step of converting the collected three-dimensional point data and live-action photo data of the transformer substation into a three-dimensional live-action point cloud map of the transformer substation comprises converting the collected three-dimensional point cloud data of the transformer substation into the three-dimensional live-action point cloud map of the transformer substation through data analysis, track analysis, coordinate conversion, point cloud generation, point cloud coloring and map distribution.
3. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the clustering and semantic segmentation of the point cloud of the three-dimensional scenic spot map comprises the steps of clustering the point cloud, segmenting the point cloud of each device, and completing automatic labeling of the semantic point cloud of the device by deep learning of the typical point cloud of the matched device; the clustering of the point clouds comprises the steps of preprocessing, filtering noise points through an outlier point cloud filtering method, and reducing the influence of the noise point clouds on equipment clustering segmentation; filtering the ground point cloud, detecting the ground through a random sampling execution algorithm, and distinguishing the detected ground data from the equipment point cloud to be clustered; and finally, accumulating the point clouds of the equipment, clustering and segmenting the point clouds to be clustered by using an European clustering algorithm based on nearest neighbor search, and matching with the point cloud information of typical equipment in a database to realize the clustering of data and complete the segmentation of semantic point clouds of the point cloud equipment.
4. A risk identification method for field operation of a transformer substation according to claim 3, wherein the random sampling execution algorithm detecting the ground comprises setting the iteration number of the algorithm to be I, and the distance error threshold to be Δt 1 The total point number of the point cloud is N, 3 points are randomly selected to form a ground to be fitted at the beginning, and three-dimensional coordinates of the 3 points are set as (X 1 ，Y 1 ，Z 1 )、(X 2 ，Y 2 ，Z 2 )、(X 3 ，Y 3 ，Z 3 ) The fitted planar model is:
Ax+By+Cz+D＝0；
wherein:
A＝(Y 2 -Y 1 )(Z 3 -Z 1 )-(Z 2 -Z 1 )(Y 3 -Y 1 )；
B＝(Z 2 -Z 1 )(X 3 -X 1 )-(X 2 -X 1 )(Z 3 -Z 1 )；
C＝(X 2 -X 1 )(Y 3 -Y 1 )-(Y 2 -Y 1 )(X 3 -X 1 )；
D＝-(AX 1 +BY 1 +CZ 1 )；
then any point (X) 0 ，Y 0 ，Z 0 ) The distance L to this plane is as follows:
when the distance L between a certain point and the assumed plane is less than or equal to delta T 1 When the model is in the model, the point is the inner point of the model; traversing other N-3 points except the initial sampling 3 points in sequence, and recording the number of the inner points of the model; randomly sampling 3 points to construct a plane model, and obtaining the number of internal points of the model according to the same method; iterating for I times according to the random sampling method; the probability of producing reasonable results increases with the increase of the iteration times; finally, voting based on the number of the internal points of each model, and selecting the ground model with the largest number of the internal points as the best fitting result; finally, the division of the ground point cloud and the point cloud with the clusters is finished, and only the point cloud with the clusters is required to be processed in the follow-up process, so that the calculation amount of the clusters is greatly reduced; when point cloud clustering is performed using the European clustering algorithm of nearest neighbor search, it is assumed that the node coordinates to be searched are (X n ，Y n ，Z n ) The coordinates of the target point are (X t ，Y t ，Z t ) And (3) making:
Δx＝X n -X t ；
Δy＝Y n -Y t ；
Δz＝Z n -Z t ；
if the 3 conditions of the I delta x I delta T, the I delta y I delta T and the I delta z I delta T are simultaneously met, calculating the Euclidean distance D between the detection point and the target point, wherein the calculation formula is as follows:
if D is less than or equal to delta T, the node is considered to be the nearest neighbor of the target point; searching all the points according to the method until all the nearest neighbor points meeting the distance error threshold are searched; recording the searched nearest neighbors of the target point in a set, and matching the point cloud data of the set with the point cloud data in a database to obtain the equipment attribute of the clustered point cloud.
5. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the steps of loading the three-dimensional point cloud map, importing the working ticket, arranging the transformer substation safety control terminal and the laser radar module comprise the steps of acquiring and automatically analyzing the working ticket, automatically reading the information of the working members and the working contents of the working ticket by accessing the PMS to automatically import the working ticket or the working ticket imported into an electronic version, automatically producing an electronic fence, and completing the arrangement of a virtual security measure; the laser radar module is arranged in such a way that a fixed-point three-dimensional laser radar module is arranged at the edge position of a working area, the laser radar realizes communication with main service equipment in a battery power supply and wireless communication mode, and the fixed-point three-dimensional laser radar module comprises a three-dimensional laser radar, an omnidirectional holder, voice intercom, voice broadcasting, wireless communication, a battery, power management and an edge processor; the transformer substation safety control terminal is a movable edge control device integrating laser radar network access, video access, data processing and control and communication on pairs, and functions of laser radar data analysis, video linkage and analysis, on-site alarm communication, remote alarm pushing and the like are realized.
6. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the clustering and semantic segmentation are carried out on the point cloud in the working area, the fusion matching of the three-dimensional point cloud map and the point cloud data comprises multi-machine calibration of a laser radar, and the three-dimensional map and the laser radar real-time data are matched; the laser radar collects real-time point cloud data of a working area and provides original data for subsequent functional analysis; the three-dimensional map and the real-time point cloud data are subjected to water level fusion matching, the calibrated laser radar data and the three-dimensional map data are in the same coordinate system, the real-time data collected by the laser point cloud can be matched with the three-dimensional map, and the real-time point cloud of the working area is displayed.
7. The risk identification method applied to field operation of a transformer substation according to claim 6, wherein the three-dimensional map and real-time point cloud data water level fusion matching comprises first constructing two point sets P and Q, wherein p= { Pi, i=1, 2,3,..k }, q= { Qi, i=1, 2,3,..n }; placing real-time points in a point set P, storing point cloud data of a three-dimensional map in a point set Q, wherein the number of the two point sets is not necessarily equal, and the conventional condition k is larger than n; the matching process is to find a rotation matrix R and a translation matrix T through rotation and translation of a coordinate system of Q, minimize the distance between corresponding points in a point set P in the point set Q, and after registration, the relationship between each point in P and Q is as follows:
P i ＝RQ i +T；
then solving a rotation matrix R and a translation matrix based on a quaternion method, and erecting a quaternion q of a unit R ＝[q 0 ，q 1 ，q 2 ，q 3 ] T WhereinAnd q 0 > 0, then the rotation matrix R is as follows:
the translation matrix is defined as q T ＝[q 4 ，q T5 ，q 6 ] T The registration vector formed by the rotation matrix and the translation matrix is After the center of gravity of the point sets P and Q is obtained and R and T are finally obtained, the Euclidean distance between the point sets P and Q can be obtained as follows:
wherein L is 2 The smaller the value is, the smaller the matching error of the two point sets is, the higher the point set identity is, and the matching of the point cloud is completed.
8. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the real-time positioning of the staff comprises analyzing the point cloud data of the staff by using a background subtraction method based on the point cloud, filtering the discrete point cloud out of the ground and equipment, and obtaining a moving point cloud cluster; clustering the point cloud clusters, obtaining the boundary of the point cloud by adopting a rolling method, then carrying out three-dimensional projection on the point cloud, and extracting point cloud characteristic information to finish target classification and point cloud skeleton information; and carrying out real-time positioning analysis on the point cloud through the point cloud type and the skeleton information to obtain real-time position data of the point cloud of the staff.
9. The risk identification method applied to the field operation of the transformer substation according to claim 1, wherein the identification of semantic point cloud, the matching of a risk identification library for deep learning, the identification and alarming of personnel risk behaviors comprise video tracking and image identification, the real-time position of the obtained personnel is combined with an image identification algorithm to perform the identification based on the face information, and the identification of the illegal behaviors are performed; based on the personnel behavior recognition of the semantic point cloud, the distance information of the personnel and the equipment is obtained by simulating the point cloud boundary through an ellipsoid model, the distance information is used for judging whether the safety distance element is installed or not, whether the behaviors of out-of-limit and wrong entering interval exist or not is judged by combining the electronic fence information, and whether the current agent has the illegal climbing behavior or not is judged according to the gravity center height; based on risk discrimination of semantic point cloud, backbone information of the semantic point cloud is fused with joint information of personnel, trend analysis of the semantic point cloud is carried out by matching the current frame point cloud with data in the previous history 10 frames, trend discrimination is carried out on the personnel, timely voice prompt is carried out on the action with potential safety hazard, and the personnel is timely reminded.
10. A risk identification system for use in field operations of a substation, comprising:
the three-dimensional scenic spot cloud scanning module is used for carrying out three-dimensional scenic spot cloud scanning on a scene of the transformer substation and collecting three-dimensional point position data and scenic photograph data of the transformer substation;
the three-dimensional real scenic spot cloud data processing module is used for converting the collected three-dimensional point data and real scenic photograph data of the transformer substation into a three-dimensional real scenic spot cloud map of the transformer substation, and clustering and semantic segmentation of the point cloud are carried out on the three-dimensional real scenic spot cloud map;
the multi-machine radar deployment and calibration module is used for loading a three-dimensional point cloud map, importing a working ticket and arranging a substation safety control terminal and a laser radar module;
the real-time point cloud acquisition module comprises a module for clustering and semantically dividing the point cloud in the working area to finish fusion matching of the three-dimensional point cloud map and the point cloud data,
the non-inductive personnel positioning module is used for monitoring personnel identity information and identifying illegal behaviors in real time through real-time positioning of the personnel;
the personnel behavior risk recognition module is used for recognizing the semantic point cloud and matching a risk recognition library for deep learning;
and the video linkage and voice alarm module is used for alarming personnel risk behaviors.
CN202310857545.3A 2023-07-13 2023-07-13 Risk identification method and system applied to field operation of transformer substation Pending CN117115728A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310857545.3A CN117115728A (en) 2023-07-13 2023-07-13 Risk identification method and system applied to field operation of transformer substation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310857545.3A CN117115728A (en) 2023-07-13 2023-07-13 Risk identification method and system applied to field operation of transformer substation

Publications (1)

Publication Number Publication Date
CN117115728A true CN117115728A (en) 2023-11-24

Family

ID=88793656

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310857545.3A Pending CN117115728A (en) 2023-07-13 2023-07-13 Risk identification method and system applied to field operation of transformer substation

Country Status (1)

Country Link
CN (1) CN117115728A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117688503A (en) * 2024-02-04 2024-03-12 国网天津市电力公司滨海供电分公司 Electricity safety inspection system based on mobile terminal

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117688503A (en) * 2024-02-04 2024-03-12 国网天津市电力公司滨海供电分公司 Electricity safety inspection system based on mobile terminal
CN117688503B (en) * 2024-02-04 2024-04-16 国网天津市电力公司滨海供电分公司 Electricity safety inspection system based on mobile terminal

Similar Documents

Publication Publication Date Title
EP3633615A1 (en) Deep learning network and average drift-based automatic vessel tracking method and system
CN103279949B (en) Based on the multi-camera parameter automatic calibration system operation method of self-align robot
CN109034018A (en) A kind of low latitude small drone method for barrier perception based on binocular vision
CN102436738A (en) Traffic monitoring device based on unmanned aerial vehicle (UAV)
CN103069434A (en) Multi-mode video event indexing
CN110602449A (en) Intelligent construction safety monitoring system method in large scene based on vision
CN102915638A (en) Surveillance video-based intelligent parking lot management system
CN111931582A (en) Image processing-based highway traffic incident detection method
CN112235537B (en) Transformer substation field operation safety early warning method
CN112738468B (en) Safety early warning method for intelligent park
CN112434566B (en) Passenger flow statistics method and device, electronic equipment and storage medium
CN114419231B (en) Traffic facility vector identification, extraction and analysis system based on point cloud data and AI technology
CN112329691A (en) Monitoring video analysis method and device, electronic equipment and storage medium
CN104933392A (en) Probabilistic people tracking using multi-view integration
CN104933436A (en) Vision-based multi-camera factory monitoring including dynamic integrity grading
CN113205116A (en) Automatic extraction and flight path planning method for unmanned aerial vehicle inspection shooting target point of power transmission line
CN117115728A (en) Risk identification method and system applied to field operation of transformer substation
CN112184773A (en) Helmet wearing detection method and system based on deep learning
CN113807240A (en) Intelligent transformer substation personnel dressing monitoring method based on uncooperative face recognition
CN115100724A (en) Multi-camera personnel state identification method
CN116846059A (en) Edge detection system for power grid inspection and monitoring
WO2011036661A1 (en) System and method for long-range surveillance of a scene and alerting of predetermined unusual activity
CN111753780A (en) Transformer substation violation detection system and violation detection method
CN112785564B (en) Pedestrian detection tracking system and method based on mechanical arm
Tarko et al. Tscan: Stationary lidar for traffic and safety studies—object detection and tracking

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination